Does it happen when you try to photograph complex images? Your glove photo has lots of detail which could be causing a buffer overflow.
You can check the images at The New Test Images - Image Compression Benchmark by making them fullscreen in your computer, and trying to take a picture.
The leaves picture seems promising
Does it happen when your phone tries to enable the flash (assuming there is one in your phone). It could be a transient power dip.
Is your phone running low on memory? I dont know how to figure this out on a palm.
Not really - there’s loads of space on the MicroSD card and internal memory - and the problem happens even after I do housekeeping to free up space
There’s no flash, but I do wonder if something about the CPU demand for compression of certain images might spike or something. I’ve got a different battery I might try for a while anyway.
This remains a possibility. I can’t reproduce the problem by taking a photo of a photo of the problem glove though - but maybe some of the contrast is lost, or some other property of the scene.
I switched the camera to save images at 1 megapixel resolution instead of 2 - the problem did not occur again (tested for a couple of weeks like that without incident).
I replaced the SD card with a new one and switched it back to 2MP and it appears stable.
So my best guess at present it’s some problem with the memory card with sustained writes beyond a certain size (certain images compressing less well than others).
My previous tinkerings with the settings to save to internal memory were misleading, as it turns out that the phone reverts to saving to the SD card on its own (I think maybe when the power is cycled or something).
I’m a bit disappointed the answer was so mundane in the end though. It would have been brilliant to have discovered a Langford Basilisk that affects JPEG compression.
Haha! I took the same kind of pictures of the same thing for years. I had a whole thingy set up on textamerica.com for them. Lost them all in the move to flickr.
I know this doesn’t help you, but I love that someone else is doing this.
I’m not an expert, but from what little I know about JPEG compression, I’m not surprised that that photo would result in a pretty large JPEG file. The entire field is “grainy” in that it’s all made up of relatively small, essentially random bits of different colors. The fleecy glove and the pebbled/sandy dirt all are made of “bumps” with pretty hard edges, from the camera’s point of view. Look closely at the glove and try to forget that as a human you know that it’s all blue. See all those little, hard-edged black dots, brown dots, light blue dots, dark blue dots, etc? JPEG compression works best on things that have smooth gradations of color, like human skin. In your glove photo, there are few places where any two adjacent pixels are of similar enough color to each other to allow JPEG to get good compression. Your later experiment of cutting down to 1 MP instead of 2 pretty much confirms this.
Try taking a picture of something like a random dot stereogram, and I’ll bet you get the same error.
I think it only kicks in at a certain level of detail. And, of course, older cameras would probably still take it, as would a film camera combined with an old scanner.
I think you’re spot on about the compression, Roadfood - it might also be that the AF on this phone works optimally at a distance of 2 feet or so, perhaps meaning that a close-up shot of something ‘pebbly’ and random may be more sharply focused (and therefore with harder boundaries between different colours) than a distance shot, even if it contains just as much randomness and grain in the subject material.
